Ensemble fire probability prediction by integrating meteorological AI models with lightweight prediction networks
7 affiliations
- 1School of Communication Engineering, Hangzhou Dianzi University, Hangzhou, Zhejiang, 310018, China
- 2School of Communication Engineering, Hangzhou Dianzi University, Hangzhou, Zhejiang, 310018, China; Pingyang Institute of Science and Technology Innovation, Wenzhou, Zhejiang, 325400, China; Corresponding author.
- 3Zhejiang Academy of Emergency Management Science and Technology, Institute of Advanced Technology, University of Science and Technology of China, Zhejiang, China
- 4Zhejiang Emergency Management Aviation Rescue Center, Zhejiang, 310061, China
- 5College of Emergency Management and Safety Engineering, North China University of Science and Technology, Tangshan, 063000, China
- 6School of Energy, Environment and Safety Engineering, China Jiliang University, Zhejiang, 310018, China
- 7Beijing University Of Technology, Digital Media Technology, Beijing, 100124, China
Abstract
Wildfires pose persistent threats to ecosystems and human safety, making timely fire-risk assessment essential for disaster prevention and resource management. However, conventional wildfire early-warning approaches are often constrained by static fire-weather indices or machine-learning models driven by reanalysis data, which cannot exploit future atmospheric fields. We therefore propose and evaluate a decoupled cascade framework that decomposes the problem into (i) an upstream AI weather model (Pangu-Weather) to generate future atmospheric states and (ii) downstream probabilistic fire classifiers. To the best of our knowledge, this is the first study to systematically compare the performance of three operational forecasting strategies (reanalysis, one-step forecast, and rolling forecast) for fire probability estimation. Using Africa as the study area, with multi-source data from 2014 to 2025, we find that one-step AI forecasts preserve the spatial fire-risk structure nearly as well as reanalysis data. Based on this dataset, we evaluated 11 downstream classifiers within a two-stage fire-occurrence and fire-intensity prediction design, while rolling forecasts suffer from degraded detection due to accumulated meteorological errors, with persistent reliability failure occurring after approximately six weeks. Cross-model consensus based on SHapley Additive exPlanations (SHAP) identifies vegetation, land-surface conditions, and meteorological variables as robust contributors. Our results demonstrate that meteorological foundation models can serve as modular and interchangeable upstream modules for operational wildfire hazard estimation, offering a scalable alternative to end-to-end black-box fire foundation models.
Related
- Cutting-edge trends in drought prediction and mitigation using ensemble learning and augmented XAI: a meteorological study
- Multi-Horizon Typhoon Wind Field Prediction via a Lightweight CNN–LSTM Network: Error Growth and Cross-Year Robustness at 6–24 h Lead Times
- Integrating Molecular Docking, Dynamics, and AI-Based ADMET Prediction to Decipher the Toxicity Profile of Isometamidium Chloride Against Animal Trypanosomiasis
- Enhancing Tropical Cyclone Prediction Through AI Weather Model Ensemble Forecasts and Multi‐Source Data Integration
- Precursors, Pathways, and State-Dependent Reliability of Long-range ENSO Prediction in the CESM2 Seasonal-to-Multiyear Large Ensemble
BibTeX
@article{wu2026ensemble,
title = {{Ensemble fire probability prediction by integrating meteorological AI models with lightweight prediction networks}},
author = {Hao Wu and Jie Luo and Xingdong Yan and Congcong Li and Jianwu Xing and Hengze Zhao and Dongmei Huang and Wei Huang and Lingxiang Wang and Bin Xu},
journal = {Ecological Indicators},
year = {2026},
month = oct,
doi = {10.1016/j.ecolind.2026.115478},
url = {https://doi.org/10.1016/j.ecolind.2026.115478},
note = {CC BY},
}
RIS
TY - JOUR
TI - Ensemble fire probability prediction by integrating meteorological AI models with lightweight prediction networks
AU - Hao Wu
AU - Jie Luo
AU - Xingdong Yan
AU - Congcong Li
AU - Jianwu Xing
AU - Hengze Zhao
AU - Dongmei Huang
AU - Wei Huang
AU - Lingxiang Wang
AU - Bin Xu
PY - 2026
DA - 2026/10/23
JO - Ecological Indicators
DO - 10.1016/j.ecolind.2026.115478
UR - https://doi.org/10.1016/j.ecolind.2026.115478
AB - Wildfires pose persistent threats to ecosystems and human safety, making timely fire-risk assessment essential for disaster prevention and resource management. However, conventional wildfire early-warning approaches are often constrained by static fire-weather indices or machine-learning models driven by reanalysis data, which cannot exploit future atmospheric fields. We therefore propose and evaluate a decoupled cascade framework that decomposes the problem into (i) an upstream AI weather model (Pangu-Weather) to generate future atmospheric states and (ii) downstream probabilistic fire classifiers. To the best of our knowledge, this is the first study to systematically compare the performance of three operational forecasting strategies (reanalysis, one-step forecast, and rolling forecast) for fire probability estimation. Using Africa as the study area, with multi-source data from 2014 to 2025, we find that one-step AI forecasts preserve the spatial fire-risk structure nearly as well as reanalysis data. Based on this dataset, we evaluated 11 downstream classifiers within a two-stage fire-occurrence and fire-intensity prediction design, while rolling forecasts suffer from degraded detection due to accumulated meteorological errors, with persistent reliability failure occurring after approximately six weeks. Cross-model consensus based on SHapley Additive exPlanations (SHAP) identifies vegetation, land-surface conditions, and meteorological variables as robust contributors. Our results demonstrate that meteorological foundation models can serve as modular and interchangeable upstream modules for operational wildfire hazard estimation, offering a scalable alternative to end-to-end black-box fire foundation models.
SN - 1470-160X
SN - 1872-7034
ER -